{"id":{"repo_id":"utc","oai_identifier":"oai:scholar.utc.edu:theses-2220"},"canonical_url":"https://search.dev.ndltd.org/etd/utc/oai:scholar.utc.edu:theses-2220","repository":{"repo_id":"utc","name":"University of Tennessee - Chattanooga","base_url":"https://scholar.utc.edu/do/oai/"},"display":{"title":"A comparative analysis of statistical and machine learning models with application in AI-powered stroke risk prediction","abstract":"Rapid detection of large vessel occlusion (LVO) is critical due to its high mortality and narrow treatment window. Although machine learning (ML) and deep learning tools show promise for LVO prediction, their clinical use is hindered by inconsistent pre-hospital data, variable LVO rates, limited interpretability, and high costs. This study introduces a hybrid neural network (HNN) that integrates classical statistical learning with neural networks to combine interpretability and structure with flexibility and regularization. The model was validated through simulations using NIHSS scores, demographics, and medical history across diverse sample sizes and LVO prevalence rates, and benchmarked against logistic regression, Naive Bayes, Decision Tree, Random Forest, and standard neural networks using sensitivity, specificity, accuracy, PPV, NPV, AUC, and ROC metrics. When applied to a large multi-center dataset from over 100 hospitals, the HNN maintained consistent performance, with sampling methods improving data balance and SHAP analysis revealing key predictors. Across simulated and real-world data, the HNN improved sensitivity by at least 20% while sustaining strong overall accuracy, demonstrating its potential as an interpretable, scalable tool for pre-hospital LVO detection to enhance clinical decision-making and improve stroke outcomes.","abstract_html":"Rapid detection of large vessel occlusion (LVO) is critical due to its high mortality and narrow treatment window. Although machine learning (ML) and deep learning tools show promise for LVO prediction, their clinical use is hindered by inconsistent pre-hospital data, variable LVO rates, limited interpretability, and high costs. This study introduces a hybrid neural network (HNN) that integrates classical statistical learning with neural networks to combine interpretability and structure with flexibility and regularization. The model was validated through simulations using NIHSS scores, demographics, and medical history across diverse sample sizes and LVO prevalence rates, and benchmarked against logistic regression, Naive Bayes, Decision Tree, Random Forest, and standard neural networks using sensitivity, specificity, accuracy, PPV, NPV, AUC, and ROC metrics. When applied to a large multi-center dataset from over 100 hospitals, the HNN maintained consistent performance, with sampling methods improving data balance and SHAP analysis revealing key predictors. Across simulated and real-world data, the HNN improved sensitivity by at least 20% while sustaining strong overall accuracy, demonstrating its potential as an interpretable, scalable tool for pre-hospital LVO detection to enhance clinical decision-making and improve stroke outcomes.","abstract_has_math":false,"creators":["McCoy, Megan"],"institution":"University of Tennessee at Chattanooga","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Gao, Lan","Barioli, Francesco; Le, Thien; Ma, Ziwei","College of Engineering and Computer Science"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-07-24T05:47:28Z","subjects":["Cerebrovascular disease--Prevention--Statistical methods","Neural networks (Computer science)","Predictive analytics"],"languages":["English","eng"],"rights":[],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[]},"links":{"outbound_url":"https://scholar.utc.edu/theses/1035","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Gao, Lan","Barioli, Francesco; Le, Thien; Ma, Ziwei","College of Engineering and Computer Science"]},{"key":"dc:creator","label":"Author","values":["McCoy, Megan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025-12-01T08:00:00Z"]},{"key":"dc:publisher","label":"Institution","values":["University of Tennessee at Chattanooga","Chattanooga (Tenn.)"]},{"key":"dc:relation","label":"Dc Relation","values":["Masters Theses and Doctoral Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["Doctoral dissertations","Text"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Cerebrovascular disease--Prevention--Statistical methods","Neural networks (Computer science)","Predictive analytics"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English","eng"]},{"key":"dc:rights","label":"Dc Rights","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://scholar.utc.edu/theses/1035"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Dept. of Mathematics","Ph. D.; A dissertation submitted to the faculty of the University of Tennessee at Chattanooga in partial fulfillment of the requirements of the degree of Doctor of Philosophy."]},{"key":"dc:description.abstract","label":"Abstract","values":["Rapid detection of large vessel occlusion (LVO) is critical due to its high mortality and narrow treatment window. Although machine learning (ML) and deep learning tools show promise for LVO prediction, their clinical use is hindered by inconsistent pre-hospital data, variable LVO rates, limited interpretability, and high costs. This study introduces a hybrid neural network (HNN) that integrates classical statistical learning with neural networks to combine interpretability and structure with flexibility and regularization. The model was validated through simulations using NIHSS scores, demographics, and medical history across diverse sample sizes and LVO prevalence rates, and benchmarked against logistic regression, Naive Bayes, Decision Tree, Random Forest, and standard neural networks using sensitivity, specificity, accuracy, PPV, NPV, AUC, and ROC metrics. When applied to a large multi-center dataset from over 100 hospitals, the HNN maintained consistent performance, with sampling methods improving data balance and SHAP analysis revealing key predictors. Across simulated and real-world data, the HNN improved sensitivity by at least 20% while sustaining strong overall accuracy, demonstrating its potential as an interpretable, scalable tool for pre-hospital LVO detection to enhance clinical decision-making and improve stroke outcomes."]},{"key":"dc:title","label":"Title","values":["A comparative analysis of statistical and machine learning models with application in AI-powered stroke risk prediction"]}]}],"canonical_facts":{"dc:contributor":["Gao, Lan","Barioli, Francesco; Le, Thien; Ma, Ziwei","College of Engineering and Computer Science"],"dc:creator":["McCoy, Megan"],"dc:date":["2025-12-01T08:00:00Z"],"dc:description":["Dept. of Mathematics","Ph. D.; A dissertation submitted to the faculty of the University of Tennessee at Chattanooga in partial fulfillment of the requirements of the degree of Doctor of Philosophy."],"dc:description.abstract":["Rapid detection of large vessel occlusion (LVO) is critical due to its high mortality and narrow treatment window. Although machine learning (ML) and deep learning tools show promise for LVO prediction, their clinical use is hindered by inconsistent pre-hospital data, variable LVO rates, limited interpretability, and high costs. This study introduces a hybrid neural network (HNN) that integrates classical statistical learning with neural networks to combine interpretability and structure with flexibility and regularization. The model was validated through simulations using NIHSS scores, demographics, and medical history across diverse sample sizes and LVO prevalence rates, and benchmarked against logistic regression, Naive Bayes, Decision Tree, Random Forest, and standard neural networks using sensitivity, specificity, accuracy, PPV, NPV, AUC, and ROC metrics. When applied to a large multi-center dataset from over 100 hospitals, the HNN maintained consistent performance, with sampling methods improving data balance and SHAP analysis revealing key predictors. Across simulated and real-world data, the HNN improved sensitivity by at least 20% while sustaining strong overall accuracy, demonstrating its potential as an interpretable, scalable tool for pre-hospital LVO detection to enhance clinical decision-making and improve stroke outcomes."],"dc:identifier":["https://scholar.utc.edu/theses/1035"],"dc:language":["English","eng"],"dc:publisher":["University of Tennessee at Chattanooga","Chattanooga (Tenn.)"],"dc:relation":["Masters Theses and Doctoral Dissertations"],"dc:rights":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:subject":["Cerebrovascular disease--Prevention--Statistical methods","Neural networks (Computer science)","Predictive analytics"],"dc:title":["A comparative analysis of statistical and machine learning models with application in AI-powered stroke risk prediction"],"dc:type":["Doctoral dissertations","Text"]},"updated_at":"2026-07-24T05:47:28Z"}